Estimating State Of Charge Of Lithium-ion Battery On Electric Vehicle Based On Adaptive Boost Algorithm

The state of charge(SOC)is an important parameter for measuring the remaining power of a power lithium-ion battery in an electric vehicle.Accurate SOC estimation can effectively prevent power lithium-ion battery to overcharge or over-discharge,which can extend battery life and ensure safe driving.Accurate and stable SOC estimation is one of the key and difficult points in the research of the power lithium-ion battery on electric vehicle.In this paper,the adaptive boost(Adaboost)regression integration algorithm is used as the framework,and the extreme learning machine(ELM)is used as the weak learner,then,the Adaboost-ELM algorithm is applied to the SOC estimation.Obtain battery voltage,current and temperature data under different temperatures and operating conditions.And using the complementary ensemble empirical mode decomposition with adaptive noise(CEEMDAN)combined with wavelet packet decomposition(WPD)and wavelet threshold denoising(WTD),two different data processing methods-CEEMDAN-WPD and CEEMDANWTD are be formed.These two methods are combined with Adaboost-ELM to build the SOC estimation model CEEMDAN-WPD-AdaboostELM(CWAELM)for power lithium-ion battery under single working condition,and the SOC estimation model CEEMDAN-WTD-AdaboostELM under various working conditions.The two models are simulated and verified by MATLAB.The results show that both models achieve an accurate estimation of the SOC.Among them,CEEMDAN-WTDAdaboost-ELM(CWAELM)model has excellent robustness and accuracy in SOC estimation of power lithium-ion battery,and has good practical use significance.